Microphysics
Microphysics copied to clipboard
AMReX-Astro
common astrophysical microphysics routines with interfaces for the different AMReX codes
The SciPy BDF solver looks quite clean and is implemented completely in python: https://github.com/scipy/scipy/blob/main/scipy/integrate/_ivp/bdf.py in the pynucastro comparisons, this seemed to perform well compared to VODE, and might make for...
It might be possible to store the Jacobian in single precision (we can still compute difference terms in double precision, and cast). It really depends on how big the Jacobian...
We have a lot of new options that need documentating including: * using both `integrator.integrate_energy` and `integrator.call_eos_in_rhs` to disable T evolution * `integrator.do_species_clip` * `unit_test.skip_initial_normalization`
PR #954 attempts to switch Jacobians within a burn. This can help, but sometimes states just fail. As an alternative, we can bring back the burn retry, with options: ```...
In `vode_dvjac.H`, we zero the Jacobian before calling the analytic Jacobian, but in the pynucastro networks, we also zero the Jacobian right when we enter the RHS routine. Note that...
Kushnir and Katz implement an equilibrium in https://ui.adsabs.harvard.edu/abs/2020MNRAS.493.5413K/abstract The idea is that if the forward and reverse rates are in balance to some tolerance then we can skip the evolution...
This paper does some nice equilibrium assumptions to make a reduced XRB / rp-process network: https://iopscience.iop.org/article/10.1086/304300/pdf
At low T, we can get really large screening factors (~ 1.e100 or so) -- these might not matter because the rates themselves are very small, but I wonder if...
In SDC, it has been observed that a zone that is originally above our T cutoff in Castro for reacting falls well below during integration, resulting in temperature (e.g. 8.e5...
Metadata
Owner
Metadata
common astrophysical microphysics routines with interfaces for the different AMReX codes